Re: [Oslib-devel] Re: OSlib ABORT 64

[Giacomo G. <gu...@li...>]

Hi all

I think the situation is more complex and it's not only a 
problem of long interruptions
 
I measured the time of a periodic_wake_up and or oneshot_wake_up functions
called by ll_timer. With the cprintf active in scheduler functions and relative scrolls 
of the video, I got 15 us on my Pentium 200 MMX... I don't understand 
where 1 ms comes from.

I think that in the shark kernel there is a bug where a cli is called without
the relative sti... so long time passes between the interrupt reactivation and
two timer interrupt come at the same time.

bye

Giacomo